KR100883520B1 - Method and apparatus for providing indoor eco-map - Google Patents

Abstract

The present invention relates to a method for rapidly obtaining an indoor environment map by using a distance measuring sensor such as a moving object and a laser sensor. The map created at this time is a grid map that divides the surrounding environment into a grid, and gives a probability that an object exists at a position corresponding to the grid. In order to prepare this, a distance measuring sensor such as a laser sensor was basically used as an object position measuring device, and a moving body equipped with an encoder was used as a vehicle. The present invention proposes a technique for creating an indoor map more quickly and accurately by using information provided from a ranging sensor and a moving object.

Indoor Map, Laser Sensor, Distance Measuring Sensor, Moving Object, Extended Kalman Filter

Description

Indoor environment mapping system and method {METHOD AND APPARATUS FOR PROVIDING INDOOR ECO-MAP}

The present invention relates to an indoor environment mapping technology, and more particularly, to an indoor environment mapping system and method suitable for accurately and quickly acquiring a structure of an indoor environment and a position of an object.

The present invention is derived from the research conducted as part of the IT new growth engine core technology development project of the Ministry of Information and Communication and the Ministry of Information and Telecommunication Research and Development. [Task Management Number: 2005-S-033-03, Assignment Name: Embedded Component Technology for URC] Development and standardization].

With the development of civilization, there are outdoor maps with various scales around the world such as cities, countries, and mountains, and more accurate and refined outdoor maps are obtained by using satellite and global positioning system (GPS). The range has also gradually expanded.

In contrast, indoor maps used in factories, public facilities, hospitals, airports, etc., despite the variety of applications, are not widely used in real life because they cannot install devices such as satellites and GPS indoors. This is because accurate location tracking (acquisition of accurate map information) is impossible and the indoor map production cost greatly increases.

On the other hand, the technology to accurately map the indoor environment and structure has a variety of applications and many research and development has been progressed. In particular, the technique of using the laser sensor to detect the position of objects in the surrounding environment and using them to create an environmental map is mainly used, which means that the laser sensor is accurate at a very narrow time interval for an arbitrary object. This is because information can be provided.

In order to create an accurate environmental map, location information acquired through a laser sensor must be determined by an accurate correlation with each other. However, if there is no device capable of measuring absolute position, such as indoor GPS, it is very difficult to accurately grasp the position information acquired by the laser sensor, and many researches and developments have been made to overcome this.

As part of such research and development, a technique called a scan matching algorithm has been dealt with in the academic world. The scan matching algorithm is a study to overcome the encoder information error of the moving object generated when the moving object equipped with the laser sensor moves from the reference position to another position by using the matching relationship between the laser sensor information.

In the paper proposed by Cox ("Blanche-an experiment in guidance and navigation of an autonomous robot vehicle," IEEE Transactions on Robotics and Automation, 1991), the "Point-to-Line matching" technique is used. Lu introduces the "Point-to-Point matching" technique through a paper ("Robot pose estimation in unknown environments by matching 2D range scans," IEEE Computer Vision and Pattern Recognition Conference (CVPR), 1994). Has been proposed. Weiss also described the "Cross Correlation Function" technique in a paper ("A map based on laser scans without geometric interpretation," In UR et al., Editor, intelligent Autonomous Systems, pages 403-407, IOS Press, 1995). Has been proposed.

The industry is approaching in various ways in consideration of the practicality of indoor environment mapping technology. Among them, a representative prior patent "Methods and apparatus for enabling a self-propelled robot to create a map of a" work area) "(KR 1997-0033627, US 5896488A1, CN 1161268 A, CN 1055772 B, CN 1161268 C, JP 3217281 B2, JP 1997-174471 A) is a technology that recognizes the environment using a robot and creates the surrounding environment. Is dealing with.

However, most of the scan matching techniques described above are focused only on automatically creating a map, so there are many exceptions that cannot be actually used, and the quality of the created map is also poor.

In addition, the environmental awareness technology of the robot does not have a detailed supplement for the case where the environmental awareness is wrong or fails.

Accordingly, an object of the present invention is to propose an indoor environment mapping system and method for quickly obtaining an environment map using location information and laser sensing information of a moving object.

In addition, the present invention is to propose an indoor environment mapping system and method that can accurately create an indoor environment map in any environment by providing the stability and user convenience of the technology.

In addition, the present invention, by applying a simple DSP technology to the system to obtain the indoor environment map corresponding to the satellite and GPS to create an indoor environment map that can quickly and accurately create the indoor structure and the location of the object to the environment map while minimizing the system configuration cost We would like to propose a system and method.

According to one aspect for solving the problems of the present invention, the moving object processing unit for operating the integrated distance value information and position information obtained from the moving object moving in the room, and applying the extended Kalman filter algorithm to predict the position information of the moving object A sensing information predicting unit predicting distance value information of the moving object using a moving model using a measuring model from the moving object position information predicting unit, the position information of the moving object predicted by the moving object position information predicting unit, and predicted measurement from the sensing information predicting unit A matching determination unit determining whether the predicted value is valid by examining whether a value is matched with an actual measured value from the moving unit processing unit; and if the predicted value is determined to be valid through the matching determination unit, A moving position information updating unit for updating position information, and the moving position information An indoor environment mapping system including an environment map preparation unit for creating an environment map according to an update result of the update unit is provided.

According to another aspect for solving the problem of the present invention, using the sensing unit for obtaining the environment information, and the moving body is equipped with the encoding unit for obtaining the rotation information of the plurality of wheels, respectively, the distance value information of the sensing unit and the encoding unit An indoor environment mapping method comprising integrated location information, the method comprising: generating location information of the moving object by the encoding unit when the moving object operates, and distance from the laser sensor to a surrounding object by the sensing unit of the moving object. Generating value information, acquiring the generated position information and distance value information, predicting the position information and the sensing information of the moving object, the position information and the sensing information of the predicted moving object, and the actual of the moving object. Determining whether the measurement information is matched, the position information and the sensor of the predicted moving object; When the information and the actual measured information of the mobile matching provides for an indoor environment map comprises: after updating the location information of the moving object which is the prediction create the indoor environment map.

According to the present invention, it is possible to quickly and accurately generate an indoor environment map for moving objects and people to recognize the indoor environment in an indoor space, for example, a public facility, a hospital, an airport, and the like. For example, when the present technology is used in the mobile robot technology, it can be used to prepare an optimal route plan by providing the mobile robot with information about the surrounding environment. In addition, the structure and features of many other indoor spaces, such as public facilities, can be easily created and used for indoor space guidance services. And if used by a construction company, it can be used to quickly and accurately create and grasp the structure of an actual building to see if the interior structure and the construction drawings match. As described above, the present invention has a great meaning in that it can be used anywhere in a technology requiring an indoor structure.

The present invention is characterized in that the structure of the surrounding environment and the position of an object can be accurately and quickly created even in a room using a laser sensor and a moving object. The core of the present invention can be classified into three types, firstly, a method of using a moving object and a laser sensor for preparing an indoor environment map, and secondly, an environmental map is quickly obtained using an extended Kalman filter. The third method is to create an environment map by point matching according to the user's acceptance of precise maps, and from this technical idea, the object of the present invention can be easily achieved.

Hereinafter, with reference to the accompanying drawings will be described in detail a preferred embodiment of the present invention.

First, Figure 1 illustrates a moving object applied to the indoor environment mapping system according to a preferred embodiment of the present invention.

As shown in FIG. 1, the movable body 100 includes a laser sensor 102, a first wheel, a second wheel, a third wheel 104a, 104b, 104c, a first encoder, Second encoders 106a, 106b.

First, the laser sensor 102 is mounted at a predetermined position of the moving object 100 and is a sensing means for acquiring surrounding environment information. At this time, in the present embodiment, for the convenience of description, the sensing means is limited to a laser sensor, but it should be noted that other distance measuring sensors, for example, an ultrasonic sensor and an infrared sensor, may be used. It will be clearer from

The first wheel, the second wheel, and the third wheel 104a, 104b, 104c are means for freely moving the movable body 100 regardless of the direction, for example, a first wheel at the front lower portion of the movable body 100. The 104a and the second wheel 104b may be mounted to the left and right, and the third wheel 104c may be mounted to the rear lower portion of the movable body 100.

In this case, the first encoder 106a and the second encoder 106b are mounted on the first wheel 104a and the second wheel 104b to calculate the distance and the direction in which the moving object 100 has moved. That is, in order to obtain the information that the first wheel 104a and the second wheel 104b are rotated, the first encoder 106a is provided on the first wheel 104a, and the second encoder 106b is provided on the second wheel 104b. Are mounted respectively.

The laser sensing information obtained through the laser sensor 102 and the encoding information obtained through the first encoder 106a and the second encoder 106b are integrated and operated by the moving object processor. This will be described in detail with reference to FIG. 2.

2 is a block diagram illustrating an indoor environment mapping system according to an aspect of the present invention, and includes a laser sensor 102, a first encoder 106a, a second encoder 106b, a moving body processing unit 200, and a moving body position. The information predicting unit 300 includes a laser sensing information predicting unit 302, a matching determination unit 304, a moving object position information updating unit 306, and an environment map preparing unit 308.

As shown in FIG. 2, the moving object processing unit 200 includes laser sensing information acquired from the laser sensor 102, that is, distance value information from the laser sensor 102 measured by the laser sensor 102 to surrounding objects. Encoding information obtained from the first encoder 106a and the second encoder 106b, that is, the first encoder 106a and the second encoder 106b mounted to the first wheel 104a and the second wheel 104b, respectively. It operates by integrating the wheel speed information measured by).

In this case, the moving object processing unit 200 manages and operates information necessary for environment map preparation in consideration of synchronization problems that may occur due to a measurement time difference between encoding information and laser sensing information. That is, in the mobile body processing unit 200, the mobile body position information converting unit 202 for converting the encoding information provided from the first encoder 106a and the second encoder 106b into the mobile body position information, and the laser sensor 102 from the laser sensor 102. A laser sensing information converting unit 204 for converting the provided surrounding environment information into laser sensing information, and a synchronizing unit 206 for synchronizing information from the moving object position information converting unit 202 and the laser sensing information converting unit 204. ) Is provided, and the synchronization information 206 is synchronized with each other and outputs the encoding information and the laser sensing information.

On the other hand, when the surrounding environment map is created using only the encoder information of the moving object 100, an error may occur in the position information of the moving object 100 due to the slipping of the wheel or the like. When the location information of the moving object 100 in which the error occurs is used as it is, it is impossible to create an accurate environment map.

Therefore, a means for correcting the generated error is required. In this embodiment, a method of quickly creating an environment map by correcting the position of the moving object having an error using the "extended Kalman filter algorithm" based on the laser sensor 102 is proposed. I would like to.

The moving object position information predicting unit 300 of FIG. 2 is a moving object position information predicting means to which the extended Kalman filter algorithm is applied.

In order to use the extended Kalman filter, a process of estimating the position of the moving object 100 using encoding information on the wheel of the moving object 100 is required, which is illustrated in FIG. 3.

번째 이동체의 위치는 다음 [수학식 1]과 같이 표현될 수 있다.As illustrated in FIG. 3, the first step for position prediction is to predict the position of the moving body of the next step from the previous step moving body position information. First in a two-dimensional plane

The position of the first moving object may be expressed as in Equation 1 below.

번째 이동체의 위치를 예측하는 과정에서 이동체가 얼마나 이동했는지를 알기 위해서는 인코더(106a)(106b)로부터의 제어 정보가 필요하다. 이러한 제어 정보는 다음 [수학식 2]와 같이 이동한 거리

와 회전한 각도

로 구성된다.And

Control information from the encoders 106a and 106b is required to know how much the moving object has moved in the process of predicting the position of the second moving object. This control information is the distance traveled as shown in [Equation 2]

And rotated angle

It consists of.

번째의 이동체 위치를 예측할 수 있다.

스텝의 이동체 위치에서 제어입력

을 받아 이동체가

만큼 회전한 후

만큼 이동하였다고 했을 때

번째의 이동체 위치는 다음 [수학식 3]과 같이 정의할 수 있다.By using this from the kinematic characteristics of the moving body as shown in FIG.

The position of the first mobile body can be predicted.

Control input at the moving body position of the step

Take a mobile

Rotate by

When you move by

The moving object position of the second can be defined as shown in [Equation 3].

는 이동체(100)의 기구학적인 모델로서, 다음 [수학식 4]와 같이 표현할 수 있다. 또한,

는 추측항법(dead reckoning)의 위치오차를 표현해 주는 것으로서, 평균이 0이고 공분산이

인 가우시안 노이즈(zeron mean Gaussian noise)로 표현된다.here

Is a kinematic model of the moving object 100, and can be expressed as Equation 4 below. Also,

Represents the positional error of dead reckoning, where the mean is zero and the covariance

It is expressed as zeon mean Gaussian noise.

는 알 수 없으 므로,

번째 추정위치

와 제어입력을 통하여 다음 [수학식 5]와 같은 결과를 예측할 수 있을 뿐이다.Actual position due to error due to floor slip

Is unknown,

First estimated position

Through the control input and and can only predict the result as shown in [Equation 5].

는 추정된 값이지만,

는 실제 제어입력으로 명확한 값이다. 그러므로 [수학식 3]을 추정된 위치

에 의한 테일러 전개(Taylor series)를 하였을 때

번째 실제위치는 다음 [수학식 6]과 같다.The error between the position predicted by Equation 5 and the actual position defined in Equation 3 is expressed in covariance form. At this time,

Is an estimated value,

Is the actual control input. Therefore, the estimated position of [Equation 3]

When Taylor developed by Taylor series

The actual position is shown in Equation 6 below.

If Equation 6 is linearized by ignoring higher order terms other than the first term, Equation 7 is expressed as follows.

는 [수학식 5]의 자코비언 행렬로 다음 [수학식 8]과같다.In [Equation 7]

Is the Jacobian matrix of Equation 5 as shown in Equation 8 below.

을 다음 [수학식 9]와 같이 구할 수 있다.Subtract the actual position value linearized in [Equation 7] and the estimated value of [Equation 5].

Can be obtained as shown in [Equation 9].

과 같다.When the error of the predicted position of [Equation 9] is expressed by a covariance matrix, the following Equation 10

Is the same as

는 선형 연산자이다. 그리고

와

은 서로 상관관계가 없으므로(un-correlated), 처음 항과 마지막 항만이 남게 되어

는 다음 [수학식 11]과 같이 전개된다.In [Equation 10]

Is a linear operator. And

Wow

Are un-correlated, leaving only the first and last terms

Is developed as shown in Equation 11 below.

은

번째 위치 오차의 공분산 행렬인

이며,

는 추측항법의 위치오차를 표현해주는 공분산 행렬

로 정리되어 최종적으로

는 최종적으로 [수학식 12]와 같이 표현될 수 있다.In Equation 11

silver

Is the covariance matrix of the second position error

Is,

Is a covariance matrix representing the positional error of the dead reckoning.

Finally organized into

Finally can be expressed as shown in [Equation 12].

번째의 이동체 위치 불확실성은

번째의 이동체 위치 불확실성이 기하학적으로

관계로 진전되며, 제어입력

및

에 의한 불확실성

가 더해진다는 물리적 의미를 담고 있다.[Equation 12] is

Mobile position uncertainty

Mobile position uncertainty is geometrically

Progress with relation, control input

And

Uncertainty by

Contains the physical meaning of being added.

On the other hand, the laser sensing information predictor 302 of FIG. 2 uses the measurement model at the position of the moving object predicted by the moving object position information predictor 300 to correct the moving object. It predicts the measured value of.

로 정의되어 있다.4 is a view for explaining the function of such a laser sensing information prediction unit 302, the local coordinate system whose position of the laser sensor 102 is based on the center of the moving object 100

It is defined as

This can be converted into a global coordinate based on the origin of the environment map, which is expressed as in Equation 13 below.

번째 정보가 탐지할 수 있는 물체의 위치를 도시한 예측 탐지 격자의 위치는 다음 [수학식 14]와 같이 정의될 수 있다.Of the laser sensor 102 in FIG.

The position of the prediction detection grid showing the position of the object that the first information can detect may be defined as in Equation 14 below.

는 레이저 센서(102)의

번째 정보가 탐지할 수 있는 물체의 위치를 나타낸 것으로, 여기서

는 레이저 센서(102)의

번째 정보가 탐지할 수 있는 전역좌표계에서의 물체의

축 위치이고,

는 전역좌표계에서의

축 위치이다.In [Equation 14]

Of the laser sensor 102

The first information is the position of the object that can be detected, where

Of the laser sensor 102

Of the object in the global coordinate system that the first information can detect

Axis position,

In the global coordinate system

The axis position.

번째 빔을 발사하였다면, 가장 가까운 물체가 존재할 격자에 의해 발사한 빔을 반사시킬 것이다. 이를 수식으로 표현하면 [수학식 15]와 같으며 이것이 측정모델이다.At this time, the laser sensor 102

If you fired the first beam, it will reflect the fired beam by the grating where the nearest object will be. This can be expressed as an equation (15), which is a measurement model.

는 레이저 센서(102)의

번째 빔의 예측 거리값이다. 여기서,

는 전역좌표계에서 레이저 센서(102)의

위치,

는 전역좌표계에서 레이저 센서(102)의

위치,

는 예측된 물체의 전역좌표계에서의

축 위치,

는 예측된 물체의 전역좌표계에서의

축 위치이다.In [Equation 15]

Of the laser sensor 102

The predicted distance value of the first beam. here,

Of the laser sensor 102 in the global coordinate system.

location,

Of the laser sensor 102 in the global coordinate system.

location,

In the global coordinate system of the predicted object

Axial position,

In the global coordinate system of the predicted object

The axis position.

번째 정보의 측정값을 다음 [수학식 16]과 같이 예측할 수 있다.By utilizing the measurement model defined in Equation 15, the laser sensor 102

The measured value of the first information can be predicted as shown in Equation 16 below.

는

번째 예측된 이동체의 위치에서 레이저 센서(102)의

번째 빔의 예측 거리값이다. 여기서,

는

번째 이동체의 위치에서 예측한

번째 이동체의 위치이며,

는 예측된

번째 이동체의 위치에서 레이저 센서(102)의

번째 정보가 탐지할 수 있는 물체의 위치를 나타낸다.In [Equation 16]

Is

Of the laser sensor 102 at the position of the first predicted moving object.

The predicted distance value of the first beam. here,

Is

Predicted from the position of the first moving object

Position of the second moving object,

Is predicted

Of the laser sensor 102 at the position of the first moving object

The first information indicates the position of the object that can be detected.

The matching determination unit 304 determines whether the predicted measurement value (Equation 16) from the laser sensing information prediction unit 302 matches the actual measurement value from the moving object processing unit 200. That is, the matching determination unit 304 determines the difference between the actual measured value of the moving object processing unit 200 and the predicted measured value of the laser sensing information predicting unit 302 according to the present embodiment, and determines whether the predicted value is valid. .

번째 정보의 측정값은 다음 [수학식 17]과 같이 정의할 수 있다.Laser sensor in Figure 4

The measured value of the first information may be defined as in Equation 17 below.

는 예측된 이동체의 위치,

는 확률격자지도에서 물체가 탐지될 위치,

는 거리값,

는 레이저 센서(102)의 오차를 의미하며, 예측된 이동체의 위치

에서 레이저 센서의

번째 정보가 빔을 발사했을 때, 확률격자지도에서 물체가 탐지될 위치

에서

의 거리값에 센서의 오차

가 더해진 값이 거리 측정값이 된다는 것이다. 레이저 센서의

번째 정보의 측정 오차는

로서 평균이 0이고 공분산이

인 가우시안 노이즈로 표시된다. 상기 [수학식 17]을 예측된 위치

에 적용하였을 때,

에 대하여 다음 [수학식 18]과 같이 테일러 전개 될 수 있다.In Equation 17

Is the predicted position of the moving object,

Where the object is to be detected on the probability grid

Is the distance value,

Denotes an error of the laser sensor 102, and predicted position of the moving object.

Of the laser sensor

Where the object will be detected on the stochastic grid when the first information fires the beam

in

Sensor error in the distance value of

The added value is the distance measurement. Of laser sensor

Error of the first information

As the mean is 0 and the covariance is

It is represented by Gaussian noise. Equation 17 is predicted position

When applied to

With respect to Taylor can be developed as shown in Equation 18.

Equation (18) is linearized by ignoring the higher order terms other than the first term.

는 [수학식 5]의 자코비언 행렬로 다음 [수학식 20]과 같다.In Equation 19

Is the Jacobian matrix of Equation 5 as shown in Equation 20 below.

는 예측된 센서의 위치에서 가장 가까운 물체가 존재하는 격자의 위치까지의 거리이다.In [Equation 20]

Is the distance from the predicted sensor position to the position of the grating where the nearest object exists.

In order to determine whether the predicted measured value is valid, an innovation is obtained as shown in Equation 21 through the difference between the predicted value of Equation 16 and the actual measured value of Equation 19.

과 같다.The covariance matrix of the innovation obtained in Equation 21 is given by Equation 22

Is the same as

는 선형 연산자이며

와

은 서로 상관관계가 없으므로 처음 항과 마지막 항만이 남게 되어

는 [수학식 23]과 같이 전개된다.In Equation 22

Is a linear operator

Wow

Are not correlated, leaving only the first and last terms

Is developed as shown in Equation 23.

은 예측된 이동체 위치의 공분산 행렬인

이며,

는 레이저 센서(102)의 측정 오차를 표현해주는 공분산 행렬

이므로 다음 [수학식 24]와 같이 정리될 수 있다.In Equation 23

Is the covariance matrix of the predicted mobile positions

Is,

Is a covariance matrix representing the measurement error of the laser sensor 102.

Because it can be summarized as follows [Equation 24].

Using the innovations of Equations 21 and 24 and the covariance matrix thereof, whether the predicted and actual measured values match well can be evaluated by the following Equation 25.

는 경계 오차 값으로서 미리 정해주는 매개변수이다. 만일, [수학식 25]를 만족한다면 매칭이 성공적임을 의미한다. 여러 개의 레이저 센서 정보를 동시에 사용하므로, [수학식 25]를 만족하는

개의 레이저 센서 정보가 있을 경우 합성이노베이션(composite innovation)은 다음 [수학식 26]과 같다.In [Equation 25]

Is a predetermined parameter as the boundary error value. If Equation 25 is satisfied, the match is successful. Since multiple laser sensor information is used at the same time, the following Equation 25

If there are two laser sensor information, synthetic innovation (composite innovation) is given by Equation 26 below.

The Jacobian value of the composite measurement model of the laser sensor 102 is also expressed by Equation 27 below.

Using Equation 27, a synthetic innovation covariance matrix can be obtained as shown in Equation 28 below.

은

개의 레이저 센서 정보에 대한 가우시안 노이즈이다.In [Equation 28]

silver

Gaussian noise for two laser sensor information.

On the other hand, the moving object position information updating unit 306 serves to update the moving object position information using the synthetic innovation (Equation 26) obtained through the matching determination unit 304.

That is, the moving object position information updating unit 306 performs the last step of the moving object correction process using the laser sensor 102, and finally updates the position of the moving object and its covariance matrix using the synthetic innovation obtained above. In this case, the well-known Kalman gain is expressed by Equation 29 below.

Using the synthetic innovation covariance matrix of [Equation 28] and Kalmangain of [Equation 29], the estimated position and the covariance of the estimated position of the moving body are updated by the following [Equation 30] and [Equation 31]. .

Finally, the environment map preparation unit 308 serves to create an environment map according to the update result of the mobile body position and the covariance matrix of the mobile body position information updating unit 306. That is, if the environment map preparation unit 308 has corrected the position of the moving object through the above processes, namely, moving object position value prediction, laser sensor value prediction, laser sensor value matching determination, estimated moving object position update, and the like, By using the location information value and the laser sensor information value to show the location of the object to the outside, a quick environment map can be created.

However, if the above algorithm is not well corrected, as shown in FIG. 5, the position of the moving object is moved manually to a position matching the laser sensor value. Using this method, environment maps can be created while correcting the position of moving objects quickly and accurately.

When creating an environment map using only the encoder information of the moving object, the position of the moving object has an error for reasons such as wheel slippage, and the extended Kalman filter algorithm was used as a means for correcting this.

In comparison, we would like to create a faster but more accurate map and suggest a way to create an environmental map at any location.

This method uses a laser sensor acquired at the position of each moving object to manually correct the position of the moving object.

Referring to FIG. 6, when the map created in the two steps is combined using two points that are matched between the map created through the laser sensor information at step k and the map generated through the laser sensor information at step k + 1, finally the correct You can create an environmental map.

와

이고, k+1 스텝에서 매칭되는 두 점이

와

이라고 하였을 때, k 스텝의 이동체 위치

로부터 k+1 스텝의 이동체 위치를 아래의 [수학식 32] 내지 [수학식 34]를 이용하여 구할 수 있으며, 이를 이용하여 상기 두 스텝(k 스텝 내지 k+1 스텝)에 걸쳐 작성된 지도를 서로 매칭시켜 정확한 지도를 작성한다.At this time, the two points matched in the k step

Wow

And two matching points in the k + 1 step

Wow

, The position of the moving body in the k step

The position of the moving object of the k + 1 step can be obtained by using the following Equations 32 to 34. Create a correct map by matching.

및

은 k번째 이동체의 레이저 센싱 정보값 중의 두 점의 위치이고,

및

은

및

과 매칭되는 k+1번째 이동체의 레이저 센싱 정보값이며,

는 k 스텝에서의 이동체의 위치,

은 k+1 스텝에서의 이동체의 위치를 각각 나타낸다.[Equation 32], [Equation 33], [Equation 34] are equations for obtaining the position of the moving object of k + 1 step, wherein

And

Is the position of two points in the laser sensing information of the k-th moving object,

And

silver

And

Is the laser sensing information of the k + 1th moving object that matches

Is the position of the moving object in k steps,

Denotes the position of the moving object in the k + 1 step, respectively.

Hereinafter, the indoor environment mapping process according to another aspect of the present invention together with the above-described configuration and functions will be described in detail with reference to the flowchart of FIG. 7.

As shown in FIG. 7, when the movable body 100 operates, the wheels 104a, 104b, 104c of the movable body 100 rotate, and the first of the wheels 104a, 104b, 104c is rotated. The positional information of the movable body 100, that is, the distance and the direction of movement of the movable body 100 by the first encoder 106a and the second encoder 106b mounted to the wheel 104a and the second wheel 104b, Is generated. At the same time, the surrounding environment information, for example, the distance value information from the laser sensor 102 to the surrounding object is generated by the laser sensor 102 of the moving object 100 (S700).

The position information of the first encoder 106a and the second encoder 106b and the distance value information of the laser sensor 102 are provided to the moving object processing unit 200, and the moving object processing unit 200 performs an information conversion and synchronization process. Going through.

The matching determination unit 304 obtains the converted and synchronized position information and distance value information of the moving object processing unit 200, that is, the actual moving position information (S702).

After that, it is determined whether to apply the Extended Kalman Filter, and when creating a fast map using the Extended Kalman Filter, the position of the moving object 100 is determined through the laser sensor-based grid map position estimation Extended Kalman Filter algorithm proposed in the present invention. If the corrected position error of the corrected moving object is large, the moving object position is corrected by manual operation to create an accurate and fast indoor environment map.

At this time, the moving object position information predictor 300 predicts the moving position information according to the present embodiment, and the laser sensing information predicting unit 302 predicts the laser sensing information of the moving object (S706). Since the moving position information and the laser sensing information prediction process have already been described with reference to FIGS. 3 and 4 and the above Equations 1 to 16, detailed descriptions thereof will be omitted.

Thereafter, in step S708, the moving object prediction information of the moving object position information predicting unit 300 and the laser sensing information predicting unit 302 and the actual measurement information of the moving object processing unit 200 are matched through the matching determination unit 304. Determine if a match is found.

If the moving object prediction information and the actual measurement information match, the process proceeds to step S710 to update the predicted moving object position information and then creates an indoor environment map. That is, the predicted moving object position information is updated through the moving object position information updating unit 306, and the updated information is provided to the environment map preparing unit 308 to create an indoor environment map showing the position of the object.

On the other hand, if the moving object prediction information and the actual measurement information does not match, the process proceeds to step S712 to correct the actual moving object position. That is, as described in FIG. 5, the position of the moving object is moved by manual manipulation to a position matching the laser sensor value.

On the other hand, if the extended Kalman filter is not used in step S704, the user can accurately correct the position of the moving object by performing manual point matching using the laser sensor measurement distance value measured at the position of each moving object (S714). . The moving object position correction method using the point matching method is as shown in FIG. 6, although the speed is less than that of the use of the extended Kalman filter, the accuracy can be guaranteed at any place.

As described above, the present invention can generate a map quickly through the extended Kalman filter algorithm, and can accurately create an indoor map at any place through the use of manual point matching.

As described above, embodiments of the present invention have been described in detail, but the present invention is not limited to these embodiments, and the features of the present invention should be understood within the spirit and scope of the present invention described in the following claims. It will also be appreciated by those skilled in the art that the invention can be operated in various modifications.

1 is a view illustrating the mounting of a moving object and a laser sensor,

2 is a block diagram illustrating a configuration of an indoor environment mapping system in which a moving object and laser sensor information are integrated and operated according to an aspect of the present invention;

3 is a diagram illustrating a moving object position prediction using encoder information of a moving object;

4 is a diagram illustrating a distance prediction value of a laser sensor;

5 is an exemplary view of manually correcting a position using a grid map and laser sensor information when a large number of errors occur during moving object position correction using the Extended Kalman Filter.

6 is a diagram illustrating a process of combining two maps by manually matching two points;

7 is a flowchart illustrating a quick indoor environment mapping process according to another aspect of the present invention.

<Brief description of the major symbols in the drawings>

102: laser sensor 106a, 106b: encoder

200: moving object processing unit 300: moving object position information prediction unit

302: laser sensing information prediction unit 304: matching determination unit

306: moving object position information updating unit 308: environment map preparation unit

Claims (16)

A method for creating an indoor environment map by integrating and operating distance value information of the sensing unit and location information of the encoding unit by using a sensing unit for acquiring surrounding environment information and a moving body equipped with an encoding unit for obtaining rotation information of a plurality of wheels, respectively. Generating location information of the mobile unit by the encoding unit when the mobile unit operates; Generating distance value information from the sensing unit to the surrounding object by the sensing unit of the moving object; Predicting location information and sensing information of the moving object after acquiring the generated location information and distance value information; Determining whether the predicted position information and sensing information of the moving object and the actual measurement information of the moving object are matched; Creating an indoor environment map after updating the location information of the predicted moving object when the position information and the sensing information of the moving object are matched with the actual measurement information of the moving object. Indoor environment mapping method comprising a. The method of claim 1, And converting and synchronizing the generated location information and distance value information. The method of claim 1, The prediction method of the location information and the sensing information of the moving object indoor map production method characterized in that the extended Kalman filter algorithm. The method of claim 3, wherein Prediction of the sensing information is the sensing unit

To predict the measured value of the first piece of information, The sensing unit

The predicted detection position of the object that the first information can detect is Equation

Defined as remind

Is the sensing unit

Location of the object the first information can be detected,

Is the sensing unit

Of the object in the global coordinate system that the first information can detect

Axial position, above

In the global coordinate system

An indoor environment mapping method, characterized in that the axis position. The method of claim 4, wherein The sensing unit

When the first beam is fired, the measurement model of the prediction detection grid where the nearest object will exist is Equation

Defined as remind

Is the sensing unit

Predicted distance value of the second beam, the

Is the sensing unit in the global coordinate system

Location, remind

Is the sensing unit in the global coordinate system

Location, remind

In the global coordinate system of the predicted object

Axial position, above

In the global coordinate system of the predicted object

An indoor environment mapping method, characterized in that the axis position. The method of claim 5, wherein The predicted using the measurement model

The predicted value of the first information is Equation

Defined as remind

Is

The sensing unit at the position of the second predicted moving object

Predicted distance value of the second beam, the

Is

Predicted from the position of the first moving object

Position value of the first mobile body,

Is predicted

Of the sensing unit at the position of the first moving object

And the second information is a position value of an object that can be detected. The method of claim 1, Creating the indoor environment map, The map created in the k step and the k + 1 step is combined by using two matching points between the map created through the sensing information of the sensing unit in the k step and the map created through the sensing information of the sensing unit in the k + 1 step. Indoor environment mapping method, characterized in that for creating an environmental map. The method of claim 7, wherein The two points matched in the k step

Wow

And the two points matched in the k + 1 step

Wow

If is, the moving body position of the k step

From the k + 1 step the moving object position, Equation

, Equation

, Equation

Is obtained from remind

And

Is the position of two points in the sensing information value of the k-th moving object,

And

silver

And

Sensing information value of the k + 1 th moving object matched with

Is the position of the moving object in the k step,

Is the position of the moving object in the k + 1 step. The method of claim 1, The indoor environment map generation method is a grid map which divides the surrounding environment into grids and gives a probability value that an object exists at a position corresponding to each grid. A moving object processing unit which integrates distance value information and location information obtained from a moving object moving indoors, A moving object position information predicting unit for predicting position information of the moving object by applying an extended Kalman filter algorithm; A sensing information predicting unit predicting distance value information of the moving object using a measurement model from the position information of the moving object predicted by the moving position information predicting unit; A matching determination unit determining whether the predicted value is valid by examining whether the predicted measured value from the sensing information predictor matches the actual measured value from the moving object processor; A moving object position information updating unit for updating the position information of the moving object if it is determined that the predicted value is valid through the matching determination unit; Environment map preparation unit for creating an environment map according to the update result of the moving position information update unit Indoor environment mapping system comprising a. The method of claim 10, The movable body, A distance measuring sensor mounted at a predetermined position of the moving object to obtain surrounding environment information; A first encoder mounted on a left wheel of the movable body and calculating a distance and a direction in which the movable body is moved; A second encoder mounted on a right wheel of the movable body to calculate a distance and a direction in which the movable body has moved Indoor environment mapping system comprising a. The method of claim 11, The distance measuring sensor is at least one of a laser sensor, an ultrasonic sensor, and an infrared sensor. The method of claim 11, And the moving object processing unit integrates and operates distance value information obtained from the distance measuring sensor and location information obtained from the first encoder and the second encoder. The method of claim 11, The moving body processing unit, A moving object position information converting unit converting encoding information provided from the first encoder and the second encoder into moving body position information; A sensing information converter configured to convert surrounding environment information provided from the distance measuring sensor into sensing information; A synchronizer for synchronizing information from the moving object position information converting unit and the sensing information converting unit; Indoor environment mapping system comprising a. The method of claim 10, And the matching determination unit determines a difference between an actual measured value of the moving object processor and a predicted measured value of the sensing information predictor to determine whether the predicted value is valid. The method of claim 10, And the indoor environment map to be created is a grid map that divides the surrounding environment into grids and gives a probability value that an object exists at a position corresponding to each grid.

Images